Your AI agents are now insiders. Who answers for them?
Security teams have spent years building programmes around one kind of insider: the employee. A new survey suggests that picture is now out of date. Software that acts on a person's behalf is turning up in incident reports, and the people being held to account for it are CISOs.
Trellix surveyed 600 CISOs at companies with more than 1,000 employees across the Americas, Europe, the Middle East and Asia-Pacific. At 95% of those companies, an AI agent or other autonomous system was linked to an insider incident in the past 12 months. Here is what the findings mean for security leaders.
Agents behave like insiders, so they are being counted as insiders
An AI agent logs in, reaches sensitive systems and acts for a person using that person's permissions. In practice, that is what an insider does. Most of the CISOs surveyed have already drawn that conclusion: 78% now include AI agents and machine identities in their definition of insider risk.
That is a meaningful shift. It means agents should be part of the same access reviews, monitoring and response plans that already apply to staff, rather than sitting in a separate category that nobody quite owns.
The visibility gap
The biggest problem the survey points to is sight. 63% of respondents say they lack sufficient visibility into how employees use AI tools. And 79% say AI makes insider incidents slower and more complicated to investigate.
Watching more closely is not always an option either. Privacy, legal and compliance limits on monitoring were the most common barrier to detection, cited by 46% of respondents. So the answer is not simply more surveillance. Security leaders need to work with legal and privacy colleagues to agree what can be monitored, and design controls around that.
Accountability is landing on the CISO
The survey also shows growing attention. 89% of companies increased their focus on insider risk over the past year. Yet 52% still lack an implemented insider-risk strategy.
Meanwhile, the responsibility is moving to one desk. An agent can run tasks on delegated credentials and make decisions that no human individually approved, which leaves it unclear who owns the outcome. 92% of CISOs say their accountability for insider incidents grew this year. 69% worry about their own liability. And 74% say their board does not clearly understand the risk.
Trellix's own CISO, David Soto, put it this way: "When AI agents operate inside your trusted boundary at machine speed, the resilient organizations will be the ones governing them like the privileged identities they are, while simultaneously building an agentic SOC to respond at the speed of the threat."
Inspired's take
The headline number will grab attention, but the more useful figure is the 74%. If three in four boards do not clearly understand the risk, CISOs are carrying accountability for something the people above them have not yet grasped.
We would suggest three practical steps:
- Treat agents as privileged identities. Give them their own credentials, limit what they can reach and review that access regularly.
- Agree ownership before an incident. Decide who owns the outcome when an agent acts, and write it down.
- Take a simple story to the board. Explain what agents can access, what could go wrong and who is responsible. Keep it short and specific.
None of this needs new technology to start. It needs clear decisions, and the CISO is well placed to push for them.
Join the conversation
How are you governing AI agents in your organisation? We would like to hear what is working and what is not. Share your view in the comments, and join fellow security leaders at our next CISO summit to compare notes.
Sources
- Help Net Security, "Most companies count AI agents as insiders, and CISOs worry about their own liability", 14 October 2026: https://www.helpnetsecurity.com/2026/10/14/ciso-ai-agent-insider-risk/
Image: padlock on a keyboard, via Unsplash (Unsplash License).
.JPG)
